Acta Cryst. (2012) D68, XXX–XXX

Supplementary materials:Exploiting structure similarity in refinement: automated NCSand target structure restraints in BUSTER

Oliver S. Smart, Thomas O. Womack, Claus Flensburg, Peter Keller, WlodekPaciorek, Andrew Sharff, Clemens Vonrhein, and Gerard Bricogne

Refinement and rebuilding of pdb entry 1det

To provide as good a target LSSR model as possible to re-solve RNAse T1-pGp basedon 5rnt data, PDB entry 1det (Ishikawa et al., 1996) was re-refined and rebuilt. 1detcrystallized in the same I23 space group as 5rnt but with a cell dimension of 88.89 Acompared to 86.47 A. Ishikawa et al. (1996) solved 1det by molecular replacement using5rnt as a search model. The original refinement (Ishikawa et al., 1996) used both X-PLOR(Brunger, 1992b) and PROLSQ (Hendrickson and Konnert, 1981). 1det has a guanosine-2’-phosphate (2’GMP) nucleotide bound and the RNAse T1 is covalently modified bycarboxylmethylation of the active site Glu58. The carboxylmethylation blocks the activesite phosphate binding site and so alters the 2’GMP binding mode compared to the“canonical” RNAse T1-2’GMP structure 1rnt (Arni et al., 1987).

The model and structure factors for 1det were obtained from the PDB (Berman et al.,2000). Re-refinement used the program BUSTER together with rebuilding using theCOOT program (Emsley et al., 2010). Although 1det has a nominal data resolution of1.8 A the BUSTER reciprocal space correlation coefficient plot showed poor data qualityabove 1.95 A resolution and below 13.5 A, so these limits were used in refinement. TheCCP4 (1994) program CAD was used to assign 5% free reflections for Rfree validation

(Brunger, 1992a). It should be noted that the free set was only used for the re-refinementand rebuild rather than throughout structure determination. Following BUSTER recom-mendations for data resolutions better than 2.0 A, hydrogen atoms were added to boththe protein and ligand using the Reduce program (Word et al., 1999). A single TLS bodyfor all atoms was used together with individual isotropic atomic B factors. StandardBUSTER restraints and default weighting schemes were used. These include Engh andHuber EH99 restraints on amino acid bond lengths and bond angles together with re-straints coupling individual temperature factors for bonded atoms. Restraint dictionariesfor the carboxymethylated glutamic acid and the 2’GMP ligand were produced using thegrade program (Smart et al., 2011) based on data obtained from the CSD database usingthe Mogul (Bruno et al., 2004) program. Following refinement the COOT program wasused to interactively rebuild the model by adjusting some of the side chain rotamers andremoving many water molecules. Table S1 shows how the refinement and rebuild lowersthe Rwork by 4% and significantly improves the MolProbity (Chen et al., 2010) validationscores.

Refinement and rebuilding shows that in the original 1det pdb structure the proteinwas in general well modeled, so only small changes were necessary. In addition the twosodium ions placed at crystal contacts in the 1det structure have good density and bindinggeometries.

s1

Figure S1: The 2’GMP ligand in 1det and the BUSTER density around it.(a) is with the original 1det pdb model, the purple arrow marks the chiral invertedatom C2’. (b) shows the better stereochemistry and fit to density for the rebuilt model.2Fo-Fc density is shown in grey and is contoured at 1.2 sigma. Fo-Fc difference densityis contoured at +3.0 sigma in green and -3.0 sigma in red. For clarity the surroundingprotein and solvent is not drawn. The figure was produced using PyMOL (DeLano, 2009).

s2

Table S1: re-refinement and rebuilding 1det.1det pdb rebuilt model

BUSTER Rwork 0.178 0.138BUSTER Rfree N/A 0.165

Number of water molecules modeled 79 47MolProbity overall score (A) 2.66 0.50MolProbity clash score 16.69 0.00MolProbity bad rotamers 4/84 0/84MolProbity Ramachandran outliers 0/101 0/100MolProbity Ramachandran favored region 96/101 99/100rms bond length deviation (A) 0.024 0.010rms bond angle deviation (degrees) 3.1 1.1MolProbity residues with bad bonds 1/101 0/100MolProbity residues with bad angles 7/101 0/100

The overall binding pose of the 2’GMP ligand in the original 1det pdb structure isreasonable with good positioning of the guanine ring, the ribose ring and the phosphategroup (Figure S1). However the stereochemistry of the 2’GMP is poor, in that thereis a chiral inversion at the 2’ carbon atom of the ribose. This inversion is clear if theribose ring is compared to the 2’GMP ideal coordinates from the ligand expo site (Fenget al., 2004) or the 2’GMP from the “canonical” RNAse structure 1rnt (Arni et al., 1987).BUSTER refinement with a grade dictionary for 2’GMP fixes the inversion problem. Inthe initial BUSTER refinement negative difference density persisted on the phosphategroup in 2’GMP ligand. This could be due to partial ligand occupancy, although alter-native explanations are radiation damage or limited disorder of the phosphate. A singlegroup occupancy variable for the 2’GMP ligand was added in the final refinement. The2’GMP occupancy refines to 0.89 and difference density around the phosphate group ismarkedly reduced (Figure S1). The real space correlation coefficient for the 2’GMP isincreased in the rebuild (Table S2). The ring pucker for the 2’GMP ribose ring providesa useful validation measure because nucleosides have well characterized pucker prefer-ences in small molecule structures (Sun et al., 2004). The ribose ring in the original 1detstructure is in the C1’-exo conformation that is rarely found in small structures (Sunet al., 2004). Refinement switches the pucker parameters (Table S2) to a favoured C2’-endo conformation (Sun et al., 2004). The Mogul strangeness score for the ring (Brunoet al., 2004) provides another indication of the ribose ring changing from an unusual toa common conformation.

The rebuilt 1det model has been deposited to the PDB and has been assigned PDBcode 3SYU.

s3

Table S2: 1det re-refinement, 2’GMP ligand statistics

1det pdb rebuilt modelBUSTER real space correlation coefficient 0.948 0.962number of chiral inversions 1 0ribose ring pseudorotation phase angle P (degrees)A 111 154ribose ring puckering amplitude νmax (degrees)A 22 40ribose ring puckerA C1’-exo C2’-endoribose Mogul ring strangeness score (degrees) 14.2 0.5A found using the PROSIT server http://cactus.nci.nih.gov/prosit/ (Sun et al.,2004)

s4

Comparison of the rebuilt-5rnt with conformational data fromLenz et al. (1993)

The rebuilt-5rnt model can be compared to conformational data quoted by Lenz et al.(1993) for the same complex solved at 1.8 A resolution but never deposited to the PDB.Table S3 shows that the conformation of the pGp ligand is similar with torsion and puckerangles with 10 degrees. An exception is for the torsion angles involving the 5’ phosphatetail, where larger differences are found. It can be noted that density is weak for the C5’atom (see Fig. 6). The C3’-endo ring pucker found is in the center of the well-favoredregion found in small molecule nucleic acid structures (Sun et al., 2004).

Table S3: pGp ligand conformationLenz et al. (1993) rebuilt

Table 2 5rnt-modelP-O5’-C5-C4’ torsion angle (degs) 165 -132O5’-C5’-C4’-C3’ torsion angle (degs) 36 62C5’-C4’-C3’-O3’ torsion angle (degs) 93 80C4’-C3’-O3’-P1 torsion angle (degs) -150 -156C5’-C4’-C3’-C2’ torsion angle (degs) -153 -163C4’-C3’-C2’-O2’ torsion angle (degs) -77 -78O4’-C1’-N9-C4 torsion angle (degs) -165 -163glycosyl bond orientation anti antiribose ring pseudorotation phase angle (degs) 13 23ribose ring pucker C3’-endo C3’-endo

Further comparison is made in Table S4 to ligand contact distances quoted by Lenzet al. (1993) for the pGp and phosphate anion found in the active site. Although theindividual contact distances differ by up to 0.5 A the same pairs of atoms are foundin all cases. Solvent molecules are the exception because Lenz et al. (1993) describe 8water molecule close to pGp and the phosphate. In contrast only one water molecule orsmall anion “107 unk” was distinguishable with the low resolution data (Fig 6.), this isequivalent to water 133 in Lenz et al. (1993).

s5

Table S4: ligand contacts distances in ALenz et al. (1993) Table 3 rebuilt 5rnt-model

pGp 105 N1 Glu 46 OE1 2.7 2.9pGp 105 N2 Glu 46 OE2 3.1 2.9pGp 105 N2 Asn 98 OE2 3.5 3.5pGp 105 N2 107 unk 3.2 3.5pGp 105 N2 107 unk 3.0 3.5pGp 105 O6 Asn 44 N 2.7 2.7pGp 105 O6 Asn 45 N 2.8 2.9pGp 105 N7 Asn 43 N 2.9 3.1pGp 105 N7 Asn 43 ND2 3.3 3.8pGp 105 O4’ His 40 NE2 3.1 3.3pGp 105 O4’ His 40 NE2 3.1 3.3pGp 105 O2P Lys 41 O 2.6 3.1pGp 105 O3P Asn 43 O 3.0 3.7PO4 106 O Glu 58 OE1 2.6 2.8PO4 106 O Glu 58 OE2 3.0 3.4PO4 106 O His 92 NE2 3.3 3.6PO4 106 O Arg 77 NE 2.8 3.1PO4 106 O Arg 77 NH2 3.0 3.0PO4 106 O 107 unk 3.1 2.7

s6

References

R. Arni, U. Heinemann, M. Maslowska, R. Tokuoka, and W. Saenger. Restrained least-squares refinement of the crystal structure of the ribonuclease T1*2’-guanylic acid com-plex at 1·9 A resolution. Acta Crystallographica Section B, 43(6):548–554, Dec 1987.doi: 10.1107/S0108768187097337.

H. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. Bhat, H. Weissig, I. Shindyalov, andP. Bourne. The Protein Data Bank. Nucleic Acids Research, 28(1):235–242, 2000. doi:10.1093/nar/28.1.235.

A. T. Brunger. Free R value: a novel statistical quantity for assessing the accuracy ofcrystal structures. Nature, 355:472–475, 1992a. doi: 10.1038/355472a0.

A. T. Brunger. X-PLOR: Version 3.1: A System for X-ray Crystallography and NMR.Yale Univ Pr, 1992b.

I. Bruno, J. Cole, M. Kessler, J. Luo, W. Motherwell, L. Purkis, B. Smith, R. Taylor,R. Cooper, S. Harris, et al. Retrieval of crystallographically-derived molecular geometryinformation. Journal of Chemical Information and Computer Sciences, 44(6):2133–2144, 2004. doi: 10.1021/ci049780b.

CCP4. The CCP4 suite: programs for protein crystallography. Acta CrystallographicaSection D, 50(5):760–763, Sep 1994. doi: 10.1107/S0907444994003112.

V. B. Chen, W. B. Arendall, III, J. J. Headd, D. A. Keedy, R. M. Immormino, G. J.Kapral, L. W. Murray, J. S. Richardson, and D. C. Richardson. MolProbity: all-atom structure validation for macromolecular crystallography. Acta CrystallographicaSection D, 66(1):12–21, Jan 2010. doi: 10.1107/S0907444909042073.

W. L. DeLano. The PyMOL Molecular Graphics System, version 1.2r1.http://www.pymol.org/, 2009.

P. Emsley, B. Lohkamp, W. G. Scott, and K. Cowtan. Features and develop-ment of Coot. Acta Crystallographica Section D, 66(4):486–501, Apr 2010. doi:10.1107/S0907444910007493.

Z. Feng, L. Chen, H. Maddula, O. Akcan, R. Oughtred, H. Berman, and J. Westbrook.Ligand Depot: a data warehouse for ligands bound to macromolecules. Bioinformatics,20(13):2153, 2004. doi: 10.1093/bioinformatics/bth214.

W. Hendrickson and J. Konnert. Stereochemically restrained crystallographic least-squares refinement of macromolecular structures. In R. Srinivasan, editor, BiomolecularStructure, Conformation, Function, and Evolution, pages 43–57. Pergamon, Oxford,1981.

K. Ishikawa, E. Suzuki, M. Tanokura, and K. Takahashi. Crystal structure of ribonucleaseT1 carboxymethylated at Glu58 in complex with 2’-GMP. Biochemistry, 35(25):8329–8334, 1996. doi: 10.1021/bi960493d.

s7

A. Lenz, H.-W. Choe, J. Granzin, U. Heinemann, and W. Saenger. Three-dimensionalstructure of the ternary complex between ribonuclease t1, guanosine 3,5-bisphosphateand inorganic phosphate at 0.19 nm resolution. European Journal of Biochemistry, 211(1-2):311–316, 1993. ISSN 1432-1033. doi: 10.1111/j.1432-1033.1993.tb19900.x.

O. S. Smart, T. O. Womack, A. Sharff, C. Flensburg, P. Keller, W. Paciorek, C. Vonrhein,and G. Bricogne. grade, version 1.1.1. http://www.globalphasing.com, 2011.

G. Sun, J. Voigt, I. Filippov, V. Marquez, and M. Nicklaus. PROSIT: pseudo-rotationalonline service and interactive tool, applied to a conformational survey of nucleosidesand nucleotides. Journal of Chemical Information and Computer Sciences, 44(5):1752–1762, 2004. doi: 10.1021/ci049881+.

J. Word, S. Lovell, T. LaBean, H. Taylor, M. Zalis, B. Presley, J. Richardson, andD. Richardson. Visualizing and quantifying molecular goodness-of-fit: small-probecontact dots with explicit hydrogen atoms1. Journal of Molecular Biology, 285(4):1711–1733, 1999. doi: 10.1006/jmbi.1998.2400.

s8